Study Guide for final

The final will cover Logic, Knowledge Representation and Planning.

Logic

• Knowledge-based agents
• The Wumpus world
• Sentences, truth, moldes, entailment
• Propositional logic, and, or, not, implication, biconditional
• Truth tables
• Logical inference
• Equivalence True in same set of models), validity (true in all models), satisfiability (true in some model)
• Resolution, conjunctive normal form, resolution theorem (a set of clauses is unstaisfiable if resolution can produce the empty clause)
• Forward and backward chaining through implications
• First-order logic, objects, relations (predicates), functions
• ASK/TELL
• Quantifiers, variables, universal, existential, equivalencies
• Equality of objects
• First-order inference, instantiation of quantified sentences, unifcation
• Resolution and conjunctive normal form in FOL

Knowledge Representation

• Ontologies
• Categories and objects
• Measurements, substances
• Situation calculus, actions, ssituations, events
• Posibility and Effect axioms, the representational frame problem
• Sucessor state axioms, fluents
• Generalized events, intervals, processes
• Semantic networks, inference about categories

Planning

• States, goals, actions, the STRIPS assumption, precondition, add and delete lists
• The blocks world
• Planning as heuristic search
• Progression and regression planning
• Partial-order planning, causal links
• Planning graphs, GRAPHPLAN
• Planning with propositional logic, SATPlan, WALKSAT

Expected skills

• Determine validity or satisfiability of a propositional sentence
• Take an English sentence and represent it in forst order logic
• Compare different first-order representations of the same English sentence
• Use eqivalences in propositional and first-order logic
• Derive a truth table for a propositional sentence
• Turn a first-order sentence into conjunctive normal form
• Use forward chaining or resolution to prove a conclusion from a KB of facts
• Represent facts involving categories, actions and events using first-order logic
• Represent a simple problem using the STRIPS formulation
• Given a STRIPS formulation of a problem, derive an action sequence for a goal